"Structuring and networking knowledge" to solve human problems
Updated by Hiroshi Komiyama on October 03, 2025, 8:00 PM JST
Hiroshi KOMIYAMA
（Platinum Initiative Network, Inc.
After serving as Professor at the University of Tokyo, Dean of the Graduate School of Engineering and Dean of the Faculty of Engineering, and President of the University of Tokyo (28th), he was appointed Chairman of the Mitsubishi Research Institute in 2009, and Chairman of the Platinum Initiative Network in 2010 (to be incorporated as a general incorporated association in 2022). Other positions include President of the STS Forum, Chairman of the Association for Super-Education, Chairman of the United Nations University Cooperation Foundation, Chairman of the International Science and Technology Foundation, and Chairman of the Heat Pump and Thermal Storage Center Foundation. He also received the Dubai Knowledge Award (2017), the Order of the Star of Solidarity of Italy (2007.) and "Information and Communication Month" Commendation from the Minister of Internal Affairs and Communications (2014), the Zaikai Award Special Prize (2016), and the Commendation for Merit in Promoting a Maritime Nation (2016), among many other national and international awards.
Humanity is currently facing two essential crises. One is the CO2 problem (e.g., climate change), which is related to the foundation of our existence on the Earth, and which is the basis of my activities as Chairman of the Platinum Network for Sustainable Development under the slogan "Carbon Neutrality by 2050. The other is the issue of generative AI, which is related to the "brain" as the foundation of knowledge. These will be the main themes of the STS Forum (International Forum on Science, Technology, and the Future of Humanity), of which I am the chairman, and will be held in Kyoto from October 5 to 7. The STS Forum will be held in Kyoto, Japan, from October 5 to 7, 2012.
The major challenge we must overcome in confronting these difficulties is the structural problem of contemporary "knowledge. The knowledge possessed by individual experts is highly specialized and is like the sharp needles of a harisenbon. Although each individual needle (specialized field) is extremely advanced, there is little understanding of the overall picture, making it difficult to solve social issues. This fragmentation is so serious even at the University of Tokyo, my former position, that even when I assumed the position of dean, I realized that it was difficult to grasp who was doing what in my own department.
I have had this sense of the challenge of "segmentation of knowledge" since I was in my master's program more than half a century ago. I had to work very hard to complete the publication of a compilation of journal articles in my area of specialization that I was simultaneously conducting with my own research. The narrowing of specialization has also affected the evaluation system for research results. In one past experiment, reviewers frequently failed to notice papers that had been resubmitted to top journals, highlighting how narrow and sophisticated specialists' specialties have become.
In response to this situation, around 1990, when I was Dean of the Faculty of Engineering at the University of Tokyo, I launched the "Structuring Knowledge Project. The "structuring of knowledge" that I advocated was to relate and organize segmented specialized knowledge in the big picture, based on the hypothesis that "all practical problems can be solved by optimally mobilizing the existing knowledge that someone else has. At that time, our goal was to construct a "virtual earth" or "virtual human being," which is equivalent to today's digital twin. However, the project hit a wall due to technical constraints and other factors.
Collaboration across disciplines is essential to solving difficult real-world problems. When I was head of the Engineering Department, I conducted a survey of research themes within the department and found, to my surprise, that about 30% of the themes were related to the human body. This suggested that the changing times have expanded the scope of engineering beyond its traditional purpose of "making things.
Knowing this reality, we promoted collaboration with medical schools, i.e., medical-industrial collaboration. Simply holding a formal meeting does not produce results if you remain confined to your "needlepoint" (specialty). I introduced a camp-style planning committee meeting, where committee members would gather overnight for lively discussions throughout the night, and by the next morning the administrative staff would compile the minutes of the meeting and re-discuss them. This was because I was keenly aware of the need for a forum where essential issues could be discussed through person-to-person communication.
And as for "structuring knowledge," I feel that the situation has fundamentally changed with the advent of generative AI. I myself am actively using generative AI. I see the most essential role of generative AI as the ability to "put together information in areas that are completely unknown to me. For example, the challenges of achieving a 2050 decarbonized society are not only scientific and technological, but also legal, organizational, institutional, administrative, and cultural. It is difficult for humans to sort through each one, but with AI, it is now possible to structure these complex groups of challenges in one fell swoop and present an overall picture of what needs to be overcome.
However, caution should be exercised at this time because generative AI is still in its developmental stages and there are issues of halucination (deceptive lies). Also, the AI developers strongly reflect their intentions (thoughts) in the system through their design. Users are required to have an attitude of understanding and using the structure and characteristics of AI proactively, rather than simply using it as a convenient tool.
I am experimenting with the use of daring to prompt instructions to the generative AI to "take your share". This is an experimental approach to adjusting the behavior of AI, which tends to pander too much to users, and to understand the relationship between humans and AI. We believe that it is more important for society as a whole to understand the structure of AI and to promote its proactive use than easy regulation.
No matter how advanced AI becomes, there will remain areas of research in deep human specialties, i.e., the "tip of the harpoon needle," that will not be replaced by AI even in 2050. Experts will continue to be needed. However, what is truly required in the age of AI is a fusion of this deep expertise and liberal arts (broad knowledge).
My definition of education is the ability to understand and discuss the "assumptions" of models specific to each field, such as economics and technology. Experts tend to get lost in their own detailed knowledge, but if the basic premise on which their expertise relies is wrong, the whole thing falls apart.
For example, in the 1995 Great Hanshin-Awaji Earthquake, highways and other structures that had been believed to be safe collapsed. The problem was not that there were errors in strength calculations, but that the "calculation assumptions" (assumed shaking) used as the basis for design were wrong. Assumptions are constantly changing and should be updated. It is dangerous in this day and age to "leave areas related to assumptions, such as economics and law, to the experts.
Our goal should be "backcasting," working backwards from an ideal vision of a decarbonized society in 2050. Individual activities alone, without an understanding of the overall structure, are not sufficient.
The solution is to increase the number of venues for fragmented experts (harisenbori needles) to discuss their assumptions and bring together their collective knowledge, rather than confining them to their own "castles". Utilize generative AI as a powerful tool to structure advanced specialized knowledge. Then, within the structure of that knowledge, a group of educated experts who can discuss the "premise of the issue" will collaborate. I am convinced that this is the way to create enormous power and solve the difficult problems toward 2050. (Hiroshi Komiyama, Chairman, Platinum Initiative Network)
New book] "Forest Circular Economy" is now on sale!
Forest Circular Economy" (edited by Hiroshi Komiyama), a book of the same title as this web media, was published by Heibonsha on August 5. The book proposes a redesign of the economy, institutions, and local communities based on the three pillars of biomass chemistry, wooden cities, and forestry innovation in the cycle of "cutting down, using, planting, and nurturing" forests. This practical book provides a point of contact between concepts and examples for those involved in policy making, social implementation, and the creation of businesses that make use of local resources.
View on Amazon
